A STUDY ON FORECASTING FINANCIAL

TIME SERIES WITH SPECIFIC REFERENCE TO

INDIAN STOCK MARKET

A THESIS

Submitted by

VASUDEVAN R D

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

FACULTY OF MANAGEMENT SCIENCES

ANNA UNIVERSITY
CHENNAI 600 025

MARCH 2018

ANNA UNIVERSITY
 CHENNAI 600 025

CERTIFICATE

The research work embodied in the present Thesis entitled “A STUDY

ON FORECASTING FINANCIAL TIME SERIES WITH SPECIFIC

REFERENCE TO INDIAN STOCK MARKET” has been carried out in the

Department of Management Studies, Kongu Engineering College, Perundurai.

The work reported herein is original and does not form part of any other thesis or

dissertation on the basis of which a degree or award was conferred on an earlier

occasion or to any other scholar.

I understand the University’s policy on plagiarism and declare that the

thesis and publications are my own work, except where specifically

acknowledged and has not been copied from other sources or been previously

submitted for award or assessment.

VASUDEVAN R D Dr. S C VETRIVEL

RESEARCH SCHOLAR SUPERVISOR
Assistant Professor
Department of
Management Studies
Kongu Engineering College
Perundurai – 638 052
 ABSTRACT

The proposition that stock prices could be well approximated by a

random walk model is a subject that has raised intense debate among capital
market regulators, investors, academics and finance professionals. Proponents
of Random Walk / Efficient Market Hypothesis argue that stock prices rapidly
and rationally adjust to reflect all available information with the result that it is
impossible to outperform the market without additional risk. On the other hand
opponents of the RW/ EM Hypothesis argue that stock price returns do maintain
trends and predictable pattern overtime. Emerging markets in general and
Indian stock markets in particular manifest characteristics that support as well as
conflict these hypotheses. Therefore, continued research on stock price
behaviour in Indian stock markets is required in order to build consensus among
the market participants.

Another phenomenon of stock markets is stock price volatility. Though

ARCH and GARCH models are apparently successful in modelling and
forecasting the volatility of financial time series data, they cannot address an
important feature called ‘leverage effect’ where the conditional variance tends to
respond asymmetrically to positive and negative shocks.

Hence, an attempt is made in this study to investigate whether Indian

stock market is weak-form efficient or not as well as estimate conditional
volatility models. The forecast accuracy of the models is evaluated in terms of
out-of-sample forecast accuracy.

The unique contribution of this study is that it analyses the above aspects
for the period January 2001 to December 2016 after dividing it into three sub-
periods, namely, pre-crisis, crisis and post-crisis periods to capture the effects of

the global financial crisis of 2008. The study investigates whether there is any
‘leverage effect’ in the Indian market during these periods. Examination of
significant differences between pre-crisis, crisis and post-crisis periods might be
useful for investors, corporate executives, portfolio managers and policy makers
in framing business policies and for the appraisal and management of present
portfolios. Therefore, the need to carry out such a research assumes greater
significance.

NIFTY data from January 2001 to December 2016 is divided into three
sub-periods, pre-crisis (Jan 2001 to Dec 2006), crisis (Jan 2007 to Dec 2010) and
post-crisis (Jan 2011 to Dec 2016) periods and the aforesaid analysis is carried
out separately for the three sub-periods to identify change in NIFTY behaviour,
if any, during the sub-periods.

The study employs descriptive statistics, correlogram, unit root test,

variance ratio test and runs test for investigating weak-form efficiency of NIFTY
over the three sub-periods. GARCH, EGARCH and TARCH models are
employed to capture volatility and leverage effect of NIFTY and develop out-of-
sample forecasts of NIFTY for 30 days at the end of each sub-period. RMSE,
MAE, MAPE and Theil inequality coefficient are employed to evaluate the
forecast accuracy of the three models.

It is inferred that NIFTY index is weak-form inefficient during pre- and

post-crisis periods but weak-form efficient during crisis period. Forecast
variations within the three ARCH family models studied are not very significant.
TARCH model, however, outperforms the other two ARCH models, while
EGARCH outperforms GARCH model in terms of forecast accuracy.
 ACKNOWLEDGEMENT

I am grateful to Dr. V Srividya, Professor, PSGIM and Dr. R

Karuppasamy, Director, Nehru Institute of management, Coimbatore, who as
members of my doctoral committee and being domain experts in finance have
contributed significantly to strengthen my research professionally with
objectivity.

Dr. R Sarvanan, Director, School of Management, Sri Krishna College

of Technology, Coimbatore was my Joint Supervisor for an year and it is
because of his guidance and persistent motivation that I was able to develop and
publish two research papers in refereed journals for which I feel extremely
obliged to him.

Dr. A G V Narayanan, Dean, Faculty of Management, EBET,

Kangayam, was my research supervisor during the first three-and-half years of
my research. He introduced me to the galaxy of experts who played a major role
in the way my research shaped-up. I am grateful to Dr. Narayanan for his
learned contribution to my research study.

Dr. S C Vetrivel, Assistant Professor (SLG), Department of Management

Studies, Kongu Engineering College, Perundurai, is my present Research
Supervisor. I truly consider it a blessing to have been groomed under the
systematic and meticulous practices of Dr. Vetrivel, during the final phase of
my research. I am indebted to Dr. Vetrivel for his personal care and
professional help that enabled early completion of my thesis.

I place on record my deep sense of appreciation to Dr. P Srinivasan,

Assistant Professor, XIME, Bangalore, who as an expert in financial
econometrics helped me on technical aspects.

VASUDEVAN R D